Using Semantic Web Ontology for Intercloud Directories and Exchanges

grotesqueoperationInternet and Web Development

Oct 21, 2013 (3 years and 7 months ago)


Using Semantic Web Ontology for Intercloud Directories and Exchanges

David Bernstein

Deepak Vij

Huawei Technologies, USA
2330 Central Expressway
Santa Clara, CA 95050

Cloud Strategy Partners, LLC
3260 Nipoma Court
San Jose, CA 95135

For submission to ICOMP'10  The 2010 International Conference on Internet Computing, Las Vegas, "V Jul 1215 2010
as part of WORLDCOMP'10 % The 2010 World Congress in Computer Science, Computer Engineering, and Applied Computing

Preacceptance draft, not for distribution

Using Semantic Web Ontology for Intercloud Directories and Exchanges

David Bernstein

Deepak Vij

Huawei Technologies, USA
2330 Central Expressway
Santa Clara, CA 95050

Cloud Strategy Partners, LLC
3260 Nipoma Court
San Jose, CA 95135

For submission to ICOMP'10  The 2010 International Conference on Internet Computing, Las Vegas, "V Jul 1215 2010
as part of WORLDCOMP'10 % The 2010 World Congress in Computer Science, Computer Engineering, and Applied Computing

Contact Author: David Bernstein. Keywords: “Cloud Computing”, “Intercloud”, “Semantic Web”, “RDF”, “Ontology”


The concept of a cloud operated by one service provider
or enterprise interoperating with a clouds operated by
another is a powerful idea. So far that is limited to use
cases where code running on one cloud explicitly
references a service on another cloud. There is no
implicit and transparent interoperability. This
interoperability should be more than cloud to cloud, it
should embody 1%to%many and many%to%many models.
Working groups have proposed building a layered set of
protocols to solve this interoperability challenge called
“Intercloud Protocols”. Instead of each cloud provider
establishing connectivity with another cloud provider in
a Point%to%Point manner resulting into n
problem, Intercloud Directories and Exchanges will
help facilitate as mediators for enabling connectivity
and collaboration among disparate cloud providers.
This paper proposes a mechanism for such mediation
utilizing a resources catalog approach, defined using
the Semantic Web Resource Definition Framework
(RDF) and a common Ontology of Cloud Computing
Resources across heterogeneous cloud providers.

1. Introduction
Cloud Computing has emerged recently as a label for
a particular type of datacenter. For the purposes of this
paper, we define Cloud Computing as a datacenter/s
1. May be hosted by anyone; an enterprise, a service
provider, or a government.
2. Implement a pool of computing resources and
services which are shared amongst subscribers.
3. Charge for resources and services using an “as
used” metered and/or capacity based model.
4. Are usually geographically distributed, in a manner
which is transparent to the subscriber (unless they
explicitly ask for visibility of that).
5. Are automated in that the provisioning and
configuration (and de-configuration and un-
provisioning) of resources and services occur on a
“self service” basis, usually programmatic request
of the subscriber, occur in an automated way with
no human operator assistance, and are delivered in
one or two orders of seconds.
6. Resources and services are delivered virtually, that
is, although they may appear to be physical (servers,
disks, network segments, etc) they are actually
virtual implementations of those on an underlying
physical infrastructure which the subscriber never
7. The physical infrastructure changes rarely. The
virtually delivered resources and services are
changing constantly.
8. Resources and services may be of a physical
metaphor (servers, disks, network segments, etc) or
they may be of an abstract metaphor (blob storage
functions, message queue functions, email functions,
multicast functions, all of which are accessed by
running of code or script to a set of API’s for these
abstract services). These may be intermixed.

Cloud Computing services as defined above are best
exemplified by the Amazon Web Services (AWS) [1][2]
or Google AppEngine [3][4]. Both of these systems
exhibit all eight characteristics as detailed above.
Various companies are beginning to offer similar
services, such as the Microsoft Azure Service [5], and
software companies such as VMware [6] and open
source projects such as UCSB Eucalyptus [7][8] are
creating software for building a cloud service.

In case 8, where the resources and services are of a
physical metaphor, the cloud is said to be exposing
“Infrastructure as a Service”, or IaaS. In the last case
described above (number 8), where the resources and
services are of an abstract metaphor, the cloud is said to
be exposing “Platform as a Service”, or PaaS. A PaaS
cloud looks like a remote, virtual, distributed
implementation of a managed code container, or
“Application Server”, similar to J2EE [9] or .NET [10].
The terms are well accepted now [11].

Use Cases and Scenarios for Cloud IaaS and PaaS
interoperability [12][13] have been detailed in the
literature along with the challenges around actually
implementing standards-based Intercloud federation and
hybrid clouds. Work detailing high level architectures
for Intercloud interoperability were proposed next
[14][15]. More recently, specific implementation
approaches for Intercloud protocols [16][17] have been
proposed, including specifically Extensible Messaging
and Presence Protocol (XMPP) [18][19] for transport,
and using Semantic Web [20] techniques such as
Resource Description Framework (RDF) [21] as a way
to specify resources.
Following that work outlining approaches for
Intercloud protocols, a detailed analysis on the
feasibility of XMPP was explored after that [22]. The
work went through considerable detail to implement
various XMPP-based control plane operations for
 Fitting XMPP into an Intercloud Topology
 Securing the XMPP conversation using TLS
 Authentication over XMPP using SAML
 Service Invocation over XMPP using IO Data
XEP, XMPP Web Services for Java (xws4j)
 RDF and SPARQL within XMPP
 XMPP Java API to a Cloud Service

The conclusion was that for each of these techniques
it found XMPP to be flexible and usable. This paper
moves to the next topic, by continuing to investigate the
blueprint set out as referenced [16][17]. We now
investigate how Cloud Computing resources can be
described, cataloged, and mediated using Semantic Web
Ontologies, implemented using RDF techniques.

2. Intercloud Topology

Cloud instances must be able to dialog with each
other. One cloud must be able to find one or more other
clouds, which for a particular interoperability scenario is
ready, willing, and able to accept an interoperability
transaction with and furthermore, exchanging whatever
subscription or usage related information which might
have been needed as a pre-cursor to the transaction.
Thus, an Intercloud Protocol for presence and
messaging needs to exist which can support the 1-to-1,
1-to-many, and many-to-many Cloud to Cloud use cases.

The vision and topology for the Intercloud we will
refer to [12][13] is as follows. At the highest level, the
analogy is with the Internet itself: in a world of TCP/IP
and the WWW, data is ubiquitous and interoperable in a
network of networks known as the “Internet”; in a world
of Cloud Computing, content, storage and computing is
ubiquitous and interoperable in a network of Clouds
known as the “Intercloud”; this is illustrated in Figure 1.

Figure 1. The Intercloud Vision

The reference topology for realizing this vision is
modeled after the public Internet infrastructure. Again,
using the generally accepted terminology
[11][12][13][14][15][18][19], there are Public Clouds,
which are analogous to ISP’s and Service Providers
offering routed IP in the Internet world. There are
Private Clouds which is simply a Cloud which an
organization builds to serve itself.
There are Intercloud Exchanges (analogous to
Internet Exchanges and Peering Points) where clouds
can interoperate, and there is an Intercloud Root,
containing services such as Naming Authority, Trust
Authority, Directory Services, and other “root”
capabilities. It is envisioned that the Intercloud root is of
course physically not a single entity, a global replicating
and hierarchical system similar to DNS [23] would be
utilized. All elements in the Intercloud topology contain
some gateway capability analogous to an Internet Router,
implementing Intercloud protocols in order to
participate in Intercloud interoperability. We call these
Intercloud Gateways. The entire topology is detailed in
Figure 2.

Comprehensive semantic descriptions of services are
essential to exploit them in their full potential. That is
discovering them dynamically, and enabling automated
service negotiation, composition and monitoring. The
semantic mechanisms currently available in service
registries such as UDDI [30] are based on taxonomies
called “tModel” [31]. tModel fails to provide the means
to achieve this, as they do not support semantic
discovery of services [32][33]. tModel supports a
construct which serves two purposes: it can serve as a
namespace for a taxonomy or as a proxy for a technical
specification that lives outside the registry. Such a
tModel construct has some intrinsic limitations, for
example classifications for the Intercloud use case can
also be defined for individual operations or their
argument types. However, this requires searching
mechanisms for services that are distinct from those for
their argument types. Likewise, tModel’s reference to an
external technical specification, as applied in UDDI also
implies that a different mechanism is required for
reasoning over service interfaces.

Although the terms “taxonomy” and “ontology” are
sometimes used interchangeably, there is a critical
difference. Taxonomy indicates only class/subclass
relationship whereas Ontology describes a domain
completely. The essential mechanisms that ontology
languages provide include their formal specification
(which allows them to be queried) and their ability to
define properties of classes. Through these properties,
very accurate descriptions of services can be defined
and services can be related to other services or resources.
We are proposing a new and improved service directory
on the lines of UDDI but based on RDF/OWL [34]
ontology framework instead of current tModel based
taxonomy framework. This catalog captures the
computing resources across all clouds in terms of
“Capabilities”, “Structural Relationships” and Policies
(Preferences and Constraints).

The following is a high level schematic of such
ontology based semantic model.

Figure 4. Cloud Computing Resources Ontology

At a very basic level, the RDF model is called a
“triple” as it consists of three parts,
Subject/Property/Object. It essentially contains one or
more “descriptions” of resources. A “description” is a
set of statements about a resource. It is structurally
similar to entity/attribute/value. Essentially, a statement
in RDF pulls resources, properties, and property values
together. Statements are typically called triples because
they include a subject (the resource), a predicate/verb
(the property), and an object (the property value or
another resource itself).

RDF allows you to define a group of things with
common characteristics called “Classes”. “Classes” are
allowed to inherit characteristics and behaviors from a
parent class. Each user-defined class is implicitly a
subclass of super class called “owl:Thing”.

The hierarchy of user-defined classes in our
proposed ontology scheme are “ResourceCapability” ￿
“CloudDomainCapability” ￿ “CloudCapability” ￿
“TierCapabil;ity” ￿ “CapabilityBundle”.

In order to demonstrate a working example, the
following is a code snippet of N-Triples [38] based
ontology semantic model instead. N-Triples & Turtle
[39] are a human-friendlier alternative to RDF/XML. N-
Triples or Turtle code, in turn, can be easily converted
to RDF/XML format using a converter tool.

The following sample shows the flow for semantic
model for cloud computing resources. Due to the large
size of the proposed semantic model for cloud
computing resources, we are unable to capture the
sample RDF code snippet in this document. In order to
demonstrate our working example, we are showing N-
Triples [38] code snippet instead.

Step 1: In our ontology example, “CloudDomain” is an
instance of class “CloudDomainCapability”. It consists
of three resources “Cloud.1”, “Cloud.2” & “Cloud.3”:



<http://cloud/domain> <

<http://cloud/domain> <
schema#label> "Cloud Computing

Step 2: “Cloud.1”, in turn, consists of tier instances
“tier.1”, “tier.2” & “tier.3”:



Step 3: Each of these cloud instances has associated
properties such as “StorageReplicationMethod”,
“InterCloudStorageAccess” etc. etc. These properties
are, in turn, used for determining if the computing
resources of a cloud provider meet the preferences &
constraints of the requesting cloud’s interest and





<> "Cloud

Step 4: Computing resources are logically grouped
together as bundles and exposed as standardized units of
provisioning and configuration to be consumed by
another cloud provider/s. These bundles are
“StorageBundle”, “ProcessingBundle” &
“NetworkBundle”. Each “Tier”, in turn, consists of
instances of resource bundles such as “StorageBundle”
etc. Each “Tier” also has its own associated properties
depicting preferences and constraints:








Step 5: “StorageBundle”, in turn, consists of resources
such as “CPU”, “CPU Cores”, “Memory” &







<> "EC2

age1> <http://www.csp/resOntology#quantity>

age1> <http://www.csp/resOntology#unit>

age1> <

age0> <http://www.csp/resOntology#quantity>

age0> <http://www.csp/resOntology#unit>

age0> <

age2> <http://www.csp/resOntology#quantity>

age2> <http://www.csp/resOntology#unit>

age2> <











6. SPARQL Query Language

SPARQL [39] (SPARQL Protocol And RDF Query
Language) is a very powerful SQL-like language for
querying and making semantic information machine
process-able. The structure and example of a SPARQL
Query is illustrated in Figure 5.

PREFIX: Prefix definition (optional)
SELECT: Result form
FROM: Data sources (optional)
WHERE: Graph pattern (=path expression)

PREFIX geo: <>
WHERE { ?X geo:hasCapital ?Y.
?Y geo:areacode ?Z }

Figure 5. Structure & Example of SPARQL Query

SPARQL provides a very powerful language for
executing very complex queries into the RDF data
which are often necessary. In our case, the following
example query applies certain Preferences and
Constraints to the resources in the computing semantics
catalog for determining if the service description on
another cloud meets the constraints of the first cloud’s
PREFIX xsd: <>
SELECT ?cld1 ?cld2 ?cld3 ?cld4 ?cld5
WHERE { ?cld1
<http://www.csp/resOntology#availabilityQuanity> ?avai
labilityQuanity .
<http://www.csp/resOntology#replicationFactor> ?replic
ationFactor .
<http://www.csp/resOntology#tierCountries> ?tierCountr
ies .
?StorageReplicationMethod .
?cld5 <http://www.csp/resOntology#
InterCloudStorageAccess > ?InterCloudStorageAccess .

FILTER ( ?availabilityQuanity = 99.999 )
FILTER ( ?replicationFactor = 5)
FILTER ( ?tierCountries = "Japan")
FILTER ( ?StorageReplicationMethod = "AMQP")
FILTER ( ?InterCloudStorageAccess = "NFS")


6.1. SPARQL Query over Hadoop

Due to very large size of “Cloud Ontology” set in the
intercloud environment, we are expecting a very large
RDF dataset. SPARQL queries against such a large RDF
dataset would be highly inefficient and slow. We believe
that such a large RDF dataset should be stored on a
Distributed File System such as HDFS (Hadoop
Distributed File System). By storing RDF dataset in
HDFS and querying through Hadoop [40] “Map-
Reduce” programming would make SPARQL queries
highly efficient and faster.

We propose that the Intercloud Exchanges will
leverage Hadoop based distributed processing for
serving SPARQL request across federated resource
catalogs hosted by Intercloud Root providers.

7. Conclusions and Future Work

We have gone into great detail to test the proposal
that Intercloud Exchanges in conjunction with Ontology
based Computing Resources Catalog and XMPP
protocol are the key components for enablement of
“Federated Cloud” environment.

As to continuing work, we are continuing to develop
the suite of Intercloud protocols. With the proposed
Intercloud Exchanges, XMPP protocol and RDF
Ontology based Resources Catalog; we should be able
to demonstrate an end-to-end comprehensive “Federated
Cloud Storage” use case for Intercloud next.

8. References

[1] Amazon Web Services at

[2] James Murty, Programming Amazon Web Services;
S3, EC2, SQS, FPS, and SimpleDB, O’Reilly Press,

[3] Google AppEngine at

[4] Eugene Ciurana, Developing with Google App
Engine, Firstpress, 2009.
[5] Microsoft Azure, at

[6] VMware VCloud Initiative at

[7] Nurmi D., Wolski R., Grzegorczyk C., Obertelli G.,
Soman S., Youseff L., Zagorodnov D., The Eucalyptus
Open%source Cloud%computing System, Proceedings of
Cloud Computing and Its Applications, Chicago, Illinois
(October 2008)
[8] Nurmi D., Wolski R., Grzegorczyk C., Obertelli G.,
Soman S., Youseff L., Zagorodnov D., Eucalyptus: A
Technical Report on an Elastic Utility Computing
Architecture Linking Your Programs to Useful Systems,
UCSB Computer Science Technical Report Number
2008-10 (August 2008)
[9] JSR 88: Java Enterprise Edition Application
Deployment at

[10] Microsoft .NET at

[11] Youseff, L. and Butrico, M. and Da Silva, D.,
Toward a unified ontology of cloud computing, GCE’08
Grid Computing Environments Workshop, 2008.
[12] Lijun Mei, W.K. Chan, T.H. Tse, A Tale of Clouds:
Paradigm Comparisons and Some Thoughts on
Research Issues, APSCC pp.464-469, 2008 IEEE Asia-
Pacific Services Computing Conference, 2008
[13] Cloud Computing Use Cases Google Group
(Public), at
, accessed March
[14] Buyya, R. and Pandey, S. and Vecchiola, C.,
Cloudbus toolkit for market%oriented cloud computing,
Proceeding of the 1st International Conference on Cloud
Computing (CloudCom ), 2009
[15] Yildiz M, Abawajy J, Ercan T., Bernoth A., A
Layered Security Approach for Cloud Computing
Infrastructure, ISPAN, pp.763-767, 10th International
Symposium on Pervasive Systems, Algorithms, and
Networks, 2009
[16] Bernstein, D., Ludvigson, E., Sankar, K., Diamond,
S., and Morrow, M., Blueprint for the Intercloud %
Protocols and Formats for Cloud Computing
Interoperability, ICIW '09. Fourth International
Conference on Internet and Web Applications and
Services, pp. 328-336, 2009
[17] Bernstein, D., Keynote 2: The Intercloud: Cloud
Interoperability at Internet Scale, NPC, pp.xiii, 2009
Sixth IFIP International Conference on Network and
Parallel Computing, 2009
[18] Extensible Messaging and Presence Protocol
(XMPP): Core, and related other RFCs at

[19] XMPP Standards Foundation at

[20] W3C Semantic Web Activity, at

[21] Resource Description Framework (RDF), at

[22] Bernstein, D., Vij, D., Using XMPP as a transport
in Intercloud Protocols, submitted to 2nd USENIX
Workshop on Hot Topics in Cloud Computing
(HotCloud '10), for publication June 2010
[23] Domain >ames – Concepts and Facilities, and
related other RFCs, at

[24] Domain >ame System Structure and Delegation, at

[25] Internet X.509 Public Key Infrastructure,
Certificate Policy and Certification Practices
Framework, at

[26] The Internet Society, at

[27] The Internet Corporation for Assigned >ames and
>umbers, at

[28] Simple Authentication and Security Layer (SASL),

[29] Security Assertion Markup Language (SAML), at

[30] OASIS UDDI Specification TC, at

[31] UDDI Registry tModels, at

[32] Paolucci, M., Kawamura T., Payne T., and Sycara
K., Importing the Semantic Web in UDDI, Web Services,
E-Business and Semantic Web Workshop, 2002.
[33] Moreau, L. and Miles, S. and Papay, J. and Decker,
K. and Payne, T., Publishing semantic descriptions of
services, First GGF Semantic Grid Workshop, held at
the Ninth Global Grid Forum, Chicago IL, USA, 2003
[34] Web Ontology Language, at

[35] Elastra, at

[36] EDML, at

[37] N-Triples, at

[38] Turtle – Terse RDF Triple Language, at

[39] SPARQL Query Language for RDF, at

[40] Hadoop, at